[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: How much is needed for sound identification ?

Yes it is, and in the text I consistently use "discrimination". The procedure was an ABX. At the time I was interested in understanding how identification processing works for everyday sounds that are very similar acoustically. This is an instance in which the title reflected the overall project, but was not accurate for the particular study.

Guillaume's work was an identification paradigm. We both used 50 ms increments, and as expected identification requires more time than discrimination. But we both concluded that the processing of everyday sounds is successful very "early".

Thanks. I'll add a clarifying note to the scanned document next to the title.
Hi Jim:

Isn't that a test of sound discrimination rather than identification?


I did a follow-up to the 1993 paper that Brian mentions and in that follow-up I used the gated stimulus paradigm.
I further chose stimulus alternatives that had been confused in the initial identification study.
After presenting the pair of sounds that had been confused, the subjects then heard one of the sounds with duration increased in 50 ms increments. Very little of the sound was needed to make the choice between which of the pair was being heard.

The paper was presented at ICAD 94, and I can scan it and send it to anyone who would want it.

The results are very task specific--the listeners could quickly adopt a strategy to focus on an initial acoustic difference between the pair and once they heard this cue, make their judgement.

However, Guillaume et al (ICAD 2004, available at http://icad.org/websiteV2.0/Conferences/ICAD2004/posters/guillaume_etal.pdf) also using a gated paradigm found that very little of the sound is needed.


-----Original Message-----
From: Sylvain Cl=E9ment [mailto:sylvain.clement@xxxxxxxxxxxxxx]
Sent: Thursday, September 13, 2007 06:49 AM
To: AUDITORY@xxxxxxxxxxxxxxx
Subject: How much is needed for sound identification ?

Dear List members,

We are currently running several experiments on environmental sounds identification vs image identification.

The literature in hearing seems to be quite poor whereas a lot of work have been done in the visual domain.

Does anybody have know studies that tried to measure how long of a sound=

>is needed to get the identification of the sound ("it's a bird") or get =

>it's super-ordonate category ("it's an animal") ?

Thanks in advance for any references.

Sylvain Cl=E9ment Neuropsychology & Auditory Cognition Team Lille, France

-- >Sylvain CLEMENT (MCF)
JE Neuropsychologie et Cognition Auditive (JE 2497)
UFR de Psychologie
BP 60149, Universite Ch. de Gaulle Lille 3
Domaine universitaire "Pont de Bois"
59 653 Villeneuve d'ascq Cedex

tel : (03 20 41) 64 42

----=_vm_0011_W8530210911_5801_1189686045 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable

Ballas (1993) measured identification of environmental sounds that were t=
runcated to 700 ms.&nbsp; He found they were quite well identified, but h=
e did not measure the time course - rather he measured reaction time, whi=
ch varied quite a bit.<br><br><font size=3D"-1"><font size=3D"3"><b>Balla=
s</b>, J. A. "Common Factors in the Identification of an Assortment of <b=

>Brief</b> Everyday <b>Sounds</b>." <b>J</b>. Exp. Psych.: Hum. Percep. =

&amp; Perf. 19 (1993): 250--26<br></font><br></font>In my environmental s=
tudies I have tried to edit sounds so that they were the briefest possibl=
e duration and still easily identifiable in pilot studies.&nbsp; I found =
quite a range of times needed - some complex events, like a bowling ball =
rolling down a lane, or a tree falling, have quite extended times courses=
(&gt; 3 s) needed to provide all the information necessary.<br><br>There=
was a study involving time course of identification of environmental sce=
nes that was presented at a conference.&nbsp; I am away from the office u=
ntil next week, but I can provide the link then.<br><br>Brian Gygi<br><br=


-- James A. Ballas, Ph.D.

Naval Research Laboratory
Code 5585
Washington, DC 20375-5337
tel: 202-404-7988
cell: 571-245-3019
fax: 202-767-1122
SIPR: ballas@xxxxxxxxxxxxxxxxxxxx

Attachment converted: Macintosh HD:smime.p7s (eApp/CSOm) (001182D1)

-- James A. Ballas, Ph.D.

Naval Research Laboratory
Code 5585
Washington, DC 20375-5337
tel: 202-404-7988
cell: 571-245-3019
fax: 202-767-1122
SIPR: ballas@xxxxxxxxxxxxxxxxxxxx

Attachment: smime.p7s
Description: S/MIME Cryptographic Signature